# Prompt Injection Discovery on AgentDojo

Source: [Prompt Injection Discovery on AgentDojo](<https://www.sota2.com/research/sota/prompt-injection-discovery-on-agentdojo>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-12T06:58:22.005Z

Evaluation focused on discovering novel prompt injection vulnerabilities within the AgentDojo environment, measured by the diversity of successful injection sources and instructions found.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Prompt Injection Discovery |
| Dataset (including any stated split) | AgentDojo |
| Metric | Source Diversity |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Source Diversity](<https://www.sota2.com/research/sota/prompt-injection-discovery-on-agentdojo.md?metric=Source%20Diversity>): Higher is better; unit: PERCENT.

- [Instruction Diversity](<https://www.sota2.com/research/sota/prompt-injection-discovery-on-agentdojo.md?metric=Instruction%20Diversity>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Source Diversity | Result date | Report paper |
| --- | --- | --- | --- |
| PI-Hunter | 84 | Not recorded | 6a2b9278d33dfb23e783af62 |
| PI-Hunter | 76 | Not recorded | 6a2b9278d33dfb23e783af62 |
| PI-Hunter | 69 | Not recorded | 6a2b9278d33dfb23e783af62 |
| Baseline | 22 | Not recorded | 6a2b9278d33dfb23e783af62 |
| Baseline | 18 | Not recorded | 6a2b9278d33dfb23e783af62 |
| Baseline | 11 | Not recorded | 6a2b9278d33dfb23e783af62 |

### PI-Hunter

Source Diversity: 84. Result ID: 6a2bae07fbc373592c297e66.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

Recorded evaluation configuration: {"Model":"Gemini-3.1-pro"}. This record does not assert independent reproduction.

### PI-Hunter

Source Diversity: 76. Result ID: 6a2bae07fbc373592c297e6a.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

Recorded evaluation configuration: {"Model":"Claude-4.6-sonnet"}. This record does not assert independent reproduction.

### PI-Hunter

Source Diversity: 69. Result ID: 6a2bae07fbc373592c297e68.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4-mini"}. This record does not assert independent reproduction.

### Baseline

Source Diversity: 22. Result ID: 6a2bae07fbc373592c297e69.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

Recorded evaluation configuration: {"Model":"Claude-4.6-sonnet"}. This record does not assert independent reproduction.

### Baseline

Source Diversity: 18. Result ID: 6a2bae07fbc373592c297e67.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4-mini"}. This record does not assert independent reproduction.

### Baseline

Source Diversity: 11. Result ID: 6a2bae07fbc373592c297e65.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b9278d33dfb23e783af62.md>)

Recorded evaluation configuration: {"Model":"Gemini-3.1-pro"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.